Propylene-propane phase equilibrium from 230 to 350 K

J. Chem. Eng. Data 1985, 30,230-233

230

Propylene-Propane Phase Equilibrium from 230 to 350 K A. Harmens Costain Petrocarbon Ltd., Manchester M22 4TB, England

Coherent tables are presented for the vapor-llquld equ#lbrlum of propylene-propane for temperatures from 230 to 350 K. The tables were calculated with a perturbed hard sphere equatkn of state, adapted specially to the saturation properties of the two substances. Measured .qulllbrlum data from the literature were used for fltting tho M a r y Interaction coeffklent. Details of the calculstkn procedure are glven. The accuracy of the equHIbrlum comporltlons is estbnated at about 2% of the smaller of the two mole fractions. I t is shown that for production of pure propylene by dloUHation, lower pressures (but not lower than about 6 bar) have some advantame over higher pressures.

Table I. Analysis of Calculated Equilibria ref Doints AAD. 9i bias. % rmsd Comparison with Underlying Empirical Data 23 9 23 19 74

7 8 11

12 total

3.08 6.18 3.09 1.97 3.17

0.95 -0.96 1.19 1.03 0.81

0.0158 0.0199 0.0138 0.0079 0.0136

Comparison with Partly Calculated Empirical Data 45 63 108

13 14

total

6.37 4.24 5.13

1.75 -0.48 0.45

0.0228 0.0188 0.0206

The parameters a and b are related to the critical constants by

Introductlon

The literature on the propylene-propane vapor-liquid equilibrium offers a somewhat disjointed picture, with the various publications being as a rule unsuitable for direct application in process design. At the same time, because of the lndustrlal importance of the system and its small relative volat#ities, there is a demand for reliable equilibrium data. I n the dedgn of a C3 distillation column the process engineer may wish to employ a water-cooled overhead condenser, which will fix his operating temperatures at around 320 K. The proximity of the critical points sets a practical limit at about 350 K. Alternatively, he may wlsh to operate at substant4Uy lower temperatures in order to exploit the larger relative volafflltks. I n that case the practical limit lies at around 230 K, below which the column pressure would become subatmoepheric. It has been the objective of this study to represent the available equlllbrlum data of this system with a single continuous calculation procedure, covering the temperatures from 230 to 350 K. Most of the empirical equilibrium data are at elevated pressures. Thermodynamic consistency analysis of such data remains somewhat controversial, because of the uncertain precision with which equations of state or generalized correlations calculate the necessary thermodynamic functions. Smith et ai. ( 7 ) have analyzed this and have used the propylene-propane system as an example. Therefore, no thermodynamic conststency analysis was attempted. A hard-sphere cubic equation of state was used, adapted to the satwation properties of the pure constituents. A single binary interaction coefficient was fiied to empirical equilibrium data, covering the entire temperature range. The calculation procedure thus obtained was used to generate the equllibrium tables. Equation of State and Mixing Rules

a = 8p2T:/p,

b = QPT,/p,

(2)

with 8, and 8,numerical factors. Pure-component data were taken from Angus et al. ( 4 ) for propylene and from Goodwin (5) for propane. The critical constants are as follows: propylene, T , = 365.57 K, p , = 46.646 bar: propane, T , = 369.80, p , = 42.42. Since eq 1 should reproduce the saturation behavior of the pure components correctly, the 8, and 3, were fitted simultaneously to vapor pressure and saturated-liquid volume data. The 8's thus obtained as functions of temperature could accurately be described by

propylene

3, = 1.10110 - 1.95859t + 0.15929t2 + 0.81806t3 Qb

= 0.12818

+ 0.28830t - 1.36975t2 + 1.04981t3

propane

3, = 1.11032 - 2.00671t + 0.17630t2 + 0.95922t3

3, = 0.12205

+ 0.33897t - 1.55702t2 + 1.29853t3

(3)

with t = 0.001 T . Equations 1 and 2 with these 8 functions (3) reproduce the underlying vapor pressures virtually exactly. For fugacity formulas, see the Appendix. For mixtures the equations were used together with the following general mixing rules:

aM =

ccx~,a, / I

bM=

ccx~!, I

/

(4)

(analogously with mole fractions y for vapor) with

A cubic equation of state as described by Iehikawa et al. (2,

3)was used

RT(2v+b)

a

p = y G --v ( v + b )

(1)

A factor of T-0.5appearing in the original attraction term was Incorporated into the parameter a . Ishikawa et ai. (2, 3) could show that in the calculation of phase equilibrium this equation

was superior to other two-parameter equations. 0021-9568/85/1730-0230$01.50/0

As shown by Mollerup ( 6 ) ,these expressions have a better foundation in the statistical theory of corresponding states than the conventional arithmetical and geometrical averages. It was found that a second interaction coefficient, in the expression for bf, could be set equal to 1.0. This stands to reason in view of the great similarity of the two molecular species. 0 1985 American Chemical Society

Journal of Chemical and Englneerlng Data, Vol. 30,

No. 2, 7985 231

Table 11. Propylene-Propane Phase Equilibrium

P, bar

x1

Y1

ai

P,bar

1.4390 1.3986 1.3542 1.3164 1.2767 1.2362 1.1931 1.1569 1.1088

4.31 4.45 4.58 4.70 4.82 4.92 5.02 5.11 5.19 5.26 5.33

1.3884 1.3556 1.3218 1.2905 1.2556 1.2223 1.1893 1.1580 1.1224

5.83 6.00 6.17 6.32 6.47 6.60 6.73 6.85 6.96 7.06 7.15

1.3452 1.3183 1.2913 1.2641 1.2367 1.2097 1.1825 1.1548 1.1277

7.70 7.92 8.12 8.32 8.50 8.68 8.84 8.99 9.14 9.27 9.39

1.3079 1.2861 1.2642 1.2420 1.2199 1.1976 1.1750 1.1526 1.1301

9.99 10.26 10.51 10.75 10.98 11.19 11.40 11.60 11.78 11.95 12.11

TemDerature = 230.0 K 0.97 1.01 1.05 1.09 1.12 1.15 1.17 1.20 1.21 1.23 1.24

0:m 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0 . m 0.1378 0.2591 0.3672 0.4674 0.5608 0.6497 0.7357 0.8223 0.9089

1.oooo

0 . m 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000

l.m

0.0000 0.1337 0.2531 0.3616 0.4625 0.5567 0.6471 0.7351 0.8224 0.9099

l.m

Temperature = 250.0 K 2.18 2.27 2.34 2.41 2.48 2.54 2.59 2.63 2.68 2.71 2.74

0 . m 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0 . m 0.1300 0.2479 0.3563 0.4573 0.5529 0.6447 0.7340 0.8220 0.9103

l.m

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0 . m 0.1269 0.2433 0.3514 0.4530 0.5495 0.6424 0.7327 0.8218 0.9105 1.0000

O.oo00 0.1241 0.2392 0.3470 0.4489 0.5462 0.6401 0.7314 0.8212 0.9104 1.0000

P , bar

1.2755 1.2578 1.2399 1.2219 1.2038 1.1856 1.1670 1.1483 1.1294

12.75 13.07 13.37 13.66 13.94 14.21 14.47 14.72 14.95 15.17 15.38

0 . m 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0.0000 0.1217 0.2356 0.3430 0.4451 0.5431 0.6377 0.7300 0.8205 0.9103 1.0000

1.2470 1.2326 1.2180 1.2033 1.1885 1.1735 1.1585 1.1429 1.1274

16.02 16.40 16.77 17.12 17.46 17.79 18.11 18.41 18.70 18.98 19.24

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0.0000 0.1195 0.2322 0.3392 0.4415 0.5399 0.6353 0.7283 0.8196 0.9100 1.0000

1.2215 1.2097 1.1978 1.1859 1.1736 1.1614 1.1490 1.1361 1.1239

19.87 20.32 20.76 21.18 21.59 21.99 22.38 22.75 23.10 23.44 23.77

1.1984 1.1887 1.1790 1.1691 1.1591 1.1489 1.1388 1.1285 1.1180

24.36 24.89 25.41 25.91 26.40 26.88 27.34 27.79 28.22 28.63 29.03

0.0000 0.1175 0.2291 0.3357 0.4380 0.5368 0.6328 0.7266 0.8186 0.9096 1.0000

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0.0000 0.1157 0.2262 0.3322 0.4346 0.5337 0.6302 0.7246 0.8174 0.9091 1.0000

ff1

1.1770 1.1690 1.1609 1.1529 1.1446 1.1363 1.1277 1.1194 1.1108

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0.0000 0.1139 0.2233 0.3289 0.4311 0.5305 0.6275 0.7225 0.8160 0.9084 1.0000

1.1567 1.1501 1.1434 1.1366 1.1299 1.1230 1.1161 1.1091 1.1020

Temperature = 330.0 K 0.0000 0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0.1122 0.2205 0.3255 0.4276 0.5271 0.6245 0.7202 0.8145 0.9076 Loo00

1.1372 1.1316 1.1260 1.1204 1.1147 1.1089 1.1032 1.0974 1.0916

Temperature = 340.0 K

Temperature = 300.0 0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

x1 Y1 Temperature = 310.0 K

Temperature = 320.0 K

Temperature = 290.0 K

Temperature = 260.0 K 3.11 3.22 3.32 3.42 3.50 3.58 3.65 3.72 3.78 3.83 3.87

O.oo00 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

a1

Temperature = 280.0 K

Temperature = 240.0 K 1.48 1.54 1.60 1.65 1.70 1.74 1.78 1.81 1.84 1.86 1.88

xi Y1 Temperature = 270.0 K

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.0000

0 . m 0.1105 0.2177 0.3220 0.4238 0.5235 0.6213 0.7176 0.8126 0.9067 1.0000

1.1177 1.1130 1.1082 1.1034 1.0986 1.0938 1.0890 1.0843 1.0794

Temperature = 350.0 K 29.56 30.18 30.79 31.38 31.96 32.52 33.07 33.60 34.12 34.62 35.10

Binary Interaction Coefficient k,/ Propylene-propane equilibrium data have been published in a number of places (7-74). The data by Hill et ai. ( 9 ) (only three points) had to be rejected as being inconsistent among themselves. Mann et ai. ( 1 0 ) reported isobaric data, without temperatures, which could not be used. Manley and Swift (73) and Laurance and Swm ( 7 4 ) measured only total pressure and speck volume of given Hqukl mixtwes and calculated the vapor composltions. That is not without risk, as Smlth et al. ( 7 ) have demonstrated. Suspicion was raised by the fact that at 310.93

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000 1.oooo

0.0000 0.1087 0.2147 0.3182 0.4197 0.5195 0.6176 0.7146 0.8104 0.9055 1.oooo

1.0975 1.0933 1.0892 1.0850 1.0811 1.0769 1.0731 1.0688 1.0651

K the two papers give quite different bubble point pressures. Therefore, these data were disallowed for the fitting of &,/, but they were used later for comparison. So fow papers remained, giving actually meawed isothermal sets of equillbrlum data for a number of temperatures. The reported purecomponent vapor pressures in general compared well with the vapor pressures used in this work: average absolute deviation 0.22% (maximum 0.40%) for propylene and 0.19% (maximum 1.12%) for propane. A computer program fitted &,/ values to the isothermal sets of empirical equilibrium data. For a given temperature it read

232 Journal of Chemical and Engineering Data, Vol. 30, No. 2, 1985

The lower part of Table I shows comparisons with the paw calculated empkical data by SWm and co-workers ( 73, 74). As was to be expected, the AAD and rmsd are somewhat larger than in the previous comparison. I t is interesting to note that the biases for the two sets of data have opposlte signs: the calculated equilibria follow an intermediate course in between the two sets. An answer to the question as to how accurately the calculated mole fractlons actually represent the true equilibrium should embody three aspects: (1) the limiting pure-component phase equilibrium is presumably represented without error: (2) the inaccuracy in the calculated mole fractions is likely to be smaller than the deviations disclosed by the tests: (3) the preponderance of the smaller mole fractions in determining the inaccuracy should be manifested. I n vlew of this, it is athated that the average absolute error In a calculated equllibrium composition is about 2 % of the smaller of the two mole fractions. The results of the tests against the 74 underlying data points also showed that almost without exception calculated x and y 1were either both too large or both too small. This being so, the numerator and the denominator of the relative volatility a1 are both affected to about the same degree by the inaccuracy in the calculated mole fractions. This makes the a1 largely insensitive to those inaccuracies.

1.5

1.4

1.3

c

-1

4

d

0

> W

2 1.2 c U

-1

W

a

1.1

I

1 1.0

I 5

I

1

I

I

I

I

10 15 20 PRESSURE BAR

25

30

Flguo 1. Relathre volatlUty a1as functbn of pressure: (-) (---) Funk and Prausnltz.

this work,

Tabulated EqulHbrla and the (a@) Diagram

In all (p,x l,y 1) data points and adjusted ku so as to mlnlmlze the sum of the relathre discrepancies between empirical x 1, x 2, yl, y2, and the corresponding calculated values, at each ( T , p ) point. So a single optimized k,, was returned for each isothermal set of data. The ku thus obtained for the data by Hanson et al. ( 8 ) at 269.54 K and by Hakuta et al. (72) at 293.25 K had to be rejected as being clearly Incompatible with the other results from the same source and far outside the general trend of k,, values. The remaining nine k4 between 228.65 and 344.26 K could be described by k,, = 0.9340

+ 0.3397t - 0.4734t2

(6)

These ku lie between 0.987 and 0.995, close to 1, as should be expected.

Tertlng Calculated Equfflbrla and Accuracy

The equuuxia which could now be calculated were first tested against the underlylng data. At each empirical ( T , p )point the four calculated equilibrium mole fractlons were compared with the corresponding empirical values. The upper part of Table I shows the results of these comparisons. They are expressed as average absdute devlatton on a percentage basis (AAD, %), bias or average deviatkm on a percentage basis (bias, %), and root mean square deviation (rmsd) in mole fractbn. “Deviation” here means calculated minus empirical mole fraction. The devlatlons are of course partly of experimental origin. AAD contains a few large lndhridual errors (up to 19%) at the extremities of the isothermal data sets,where the small propylene or propane mole fracths have magnmed the percentage deviations. Since errors in the middle of the concentration range do not Contribute materially to the bias (contrlbutions for propylene and propane largely cancel out),the slightly posklve bias indicates that the smaHer concentrations tend to be calculated slightly too high.

The computation procedure was incorporated into a bubble point program to generate the equilibrium data and relative volatilities a1shown in Table 11. The propylene-propane equilibrium has been analyzed before, by Funk and Prausnitz (75), using the composite thermodynamic treatment of Prausnb and Chueh (76). They presented their results in an ( a l p )diagram showing llnes for constant x 1. A similar diagram was prepared on the basis of our equation of state procedure (Figure 1). I t shows curves for five x 1values, and three corresponding curves from ref 75. At low pressures there is a measure of agreement, but toward higher pressures the Funk-Prausnltz curves run lncreaslngly too high. This is probably caused by insufficient refinement of the Prausnltz-Chueh procedure, which uses constant B, and Bb and (for this system) a k, = 1.0. The present x = 0.99 curve shows a maximum at about 6 bar. At low pressures lt falls toward a1= 1.0. The reason for this is that propane has only a slightly lower vapor pressure than propylene whHe the liquid mixtures are almost bl:at constant T the total pressure at the propylene end is almost constant. Table I1 confkms this. Since at the propylene end Raoult‘s law approximately holds for propylene, yllxl equals about 1 and consequently cyl approaches I.Toward higher pressures the present curve is noticeably steeper than the Funk-Prausnitz curve. This indicates that for producing pure propylene by distillation it is advantageous to operate at lower pressures, but not below about 6 bar.

Appendlx Fugacity Fomlas for Eq 1 . For a single substance

Z- 1 - In Z ( 7 )

J. Chem. Eng. Data 1985,30,233-234

For a component i in a mixture

critical mixture

C

M

(L ) +

233

in cp, = 2 in 2 ~ b-

(K 2 -

Roghtry

&Xb-

I n this formula a and b equal a

,,and b

g) -

V X

Y Z a1

cp Q,, f i b

in z (8)

of eq 4.

parameters in equation of state (1) binary interaction coefficient total pressure, bar gas constant (83.1448 bar.cm3/moi-K) absolute temperature, K 0.001 T molal volume, cm3/moi mole fraction in iiquM mole fraction in vapor compressibility factor pw IRT relative volatility y ,x,/x ly2 fugacity coefficient numerical factors In a and b , eq 2 and 3

Subscripts

1, 2 1, i

Propylene, 115-07-1; propane, 74-98-6.

Literature Cited

--ry a ,b k P R T t

No.

refer to propylene and propane, respectively refer to any of the components

(I) Smith, E. D.; Muthu, 0.;Dewan, A.; Gierlach. M. J . fhys. Chem. Ref. Data 1982, I I , 941. (2) Ishikawa, T.; Chung, W. K.; Lu, E. C. Y. A I W J . 1980. 26, 372. (3) Ishkawa, T.; Chung, W. K.; Lu, B. C. Y. A&. Cryog. Eng. 1980, 25, 671. (4) Angus, S.; Armstrong, E.: de Reuck, K. M. “International Thecmodynamlc Tabbe of the Fluld State, 7-Propylene”; International Union of Pure and AppW Chembtty; Oxford, 1980. (5) Goodwln, R. D. ”Provldonal Thermodynamic Functions of Propane from 85 to 700 K at Pressures to 700 bar”; US. National Bureau of Standards: Wadrlngton, DC, July 1977; NBSIR-77-860. (8) Mollerup, J., presented at the 71st Annual Meeting of the American Institute of Chemkal Engineers Mlami, FL, Nov 1978. (7) Reamer, H. H.; Sage, E. H. Ind. Eng. Chem. 1951, 43, 1628. (8) Hanson, G. H.; Hogan, R. J.; Nelson, W. T.: Cines, M. R. Ind. Eng. Cl”.1952, 44, 804. (9) Hili, A. E.; McCormlck, R. H.; Barton, P.; Fenske, M. R. A I C M J . 1982, 8 , 681. (10) Mann. A. N.; Pardee, W. A.; Smyth, R. W. J . Chem. Eng. Data 1983, 8, 499. (11) Hlrata, M.; Hakuta, T.; On&, T. Int. (3”.Eng. 1968, 8 . 175. (12) Hakuta, T.; Nagahama, K.; Hlrata, M. Bull. Jpn. Pet. Inst. 1989, 1 1 , 10. (13) Manby, D. E.; SwM, (3. W. J . Chem. Eng. Data 1971, 16, 301. (14) Laurence, D. R.; SwM, 0. W. J . Chem. Eng. Data 1972, 17, 333. (15) Funk, E. W.; Prausnltz. J. M. A I W J . 1971. 17, 254. (18) Rausnitz, J. M.; Chwh, P. L. “Computer Calculations for High Pressure Vapor-Liquid EquHlbria”; Prentice-Hall: Englewood CIHls, NJ, 1968.

Recehred for review May 21, 1964. Accepted October 9, 1984.

Vapor-Liquid Equilibrium in Aqueous Solutions of Various Glycols and Poly( ethylene glycols), 3. Poly(ethy1ene glycols) Mordechay Herskowltz and Moshe Gottlleb’ Chemical Engineering Department, Ben-Gurion University of the Negev, Beer Sheva, 84 105 Israel

The activity of water In sdutionr of poly(ethyhme glycols) (PEGS, molecular weights 200, 600, 1500, 6000) was measured over a wide range of weigM fractions at 293.1, 313.1, and 993.1 K. The data were obtained by an isopktk method. A comparison between the measured actlvlties and predicted values by the UNIFAC method gives a good agreement for PEG 200 solutions only.

isopiestic method to measure the water activity in aqueous solutions of poly(ethyiene glycols) (molecular weights 20020000) at 298 K. This study was limited to relatively dilute solutions (polymer weight fraction less than 0.5). I n this work we have enlarged the body of available data on the activity of aqueous PEG solutions by obtaining data for polymers in the molecular weight range of 200-6000 over the entire concentration range.

Introduction

Experimental Section

Poty(ethylene glycols) are polymers that And a wide range of industrialapplications due to thek hlgh solubility in water. They are commonly used in the pharmaceutical industry as excipients in drug formulation, as svfac+acthre agents in water treatment, as fiber-forming aids in the textile industry, in the manufacture of lubricants and mold release agents, and recently as reactive roiebies in the preparationof hydrophk polyurethane networks for medical purposes ( 1 , 2). The properties of the aqueous sdutrons of poIy(ethylene glycols) have been extensively studied by various methods (3-6). Malcolm and Rowiinson (7) measured the vapor pressures of aqueous solutions of poiy(ethyiene glycol) of molecular weights 300,3000. and 5000 at 303-338 K. The water actMty was calculated from the data. Adamcova (8) employed an

The isopiestic apparatus employed in this study is described in detail elsewhere (9). The reference solute was l i i u m chloride, manufactured by Merck Co. Its purity analyzed by atomic absorption and titration was better than 99.8%. The poly(ethylene glycols), manufactured by BDH, were used as supplied. After the salt and the polymers were dried at 120 OC, their water content measured by Karl Flscher analysis was less than 0.3 % and 0.7 % , respectively.

0021-9568/85/1730-0233$01.50/0

Resuits and Dlrcurrlon The water activity was Calculated from the expression

in a , = -vpr~,/55.51 where v,, m,, and #IJ

(1)

,are the number of ions, the molality, and

0 1985 American Chemlcal Society